References & Citations
Statistics > Computation
Title: Approximate Bayesian Computations to fit and compare insurance loss models
(Submitted on 8 Jul 2020 (v1), last revised 1 May 2021 (this version, v2))
Abstract: Approximate Bayesian Computation (ABC) is a statistical learning technique to calibrate and select models by comparing observed data to simulated data. This technique bypasses the use of the likelihood and requires only the ability to generate synthetic data from the models of interest. We apply ABC to fit and compare insurance loss models using aggregated data. A state-of-the-art ABC implementation in Python is proposed. It uses sequential Monte Carlo to sample from the posterior distribution and the Wasserstein distance to compare the observed and synthetic data.
Submission history
From: Patrick J. Laub [view email][v1] Wed, 8 Jul 2020 00:41:34 GMT (113kb)
[v2] Sat, 1 May 2021 03:41:40 GMT (185kb)
Link back to: arXiv, form interface, contact.